Plant Fluid State Estimation Using ML Instead of Real-Time CFD
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies, such as computational fluid dynamics (CFD), require enormous computational resources and are unable to estimate the state of fluids in real-time within plant operations.
Innovation Solution
A fluid state estimation system that includes a learning device to acquire and learn an estimation model using machine learning, and an estimation device that uses this model to quickly and accurately estimate fluid states within various components and environments of a plant.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamics (CFD) is used to estimate fluid state with high accuracy, then measurement precision is improved, but productivity deteriorates due to enormous calculation requirements preventing real-time estimation
Solution Approach 1:
The system performs preliminary actions by collecting fluid state data and operating condition data during plant operation, then uses a learning device to train an estimation model in advance. This pre-trained model can subsequently provide real-time fluid state estimates without requiring intensive CFD calculations during actual operation, thus resolving the contradiction between accuracy and real-time capability.
Solution Approach 2:
The invention creates a simplified copy of the complex CFD-based fluid state estimation system by training a machine learning model on CFD data and operational data. This copied model replicates the accuracy of CFD simulations but executes much faster, enabling real-time estimation while maintaining measurement precision.
2Measurement precision
If CFD calculations are performed to obtain accurate fluid state information, then measurement precision is improved, but loss of time increases due to the enormous computational resources required
Solution Approach 1:
The system performs the computationally intensive work in advance by training the estimation model during periods when time is not critical. Once trained, the model can provide accurate fluid state information instantly during operation, eliminating the time loss associated with real-time CFD calculations while maintaining measurement precision.
Solution Approach 2:
The system dynamically switches between two modes: an offline training phase using CFD data to build the model, and an online estimation phase using the trained model for rapid predictions. This dynamic approach allows the system to achieve both high accuracy and fast response times by performing heavy calculations only when necessary.
Data Source
AI summary
A fluid state estimation system includes: a learning device that learns an estimation model for estimating a state of a fluid in at least any one of an inside of a component of a plant, an outside of the component of the plant and an inside of a building of the plant, an outside of the building of the plant and an inside of a site of the plant, and a periphery of an outside of the site of the plant; and an estimation device that estimates a state of a fluid using the estimation model learned by the learning device. The estimation model gets a value of an input variable, and outputs a value of fluid state information representing a state of a fluid.